Implementation Of Content-based The County Party-governmetn Conference Video Retrieval System

With the rapid development of networks and information technology, Party-government conference videos contains lots of important information about the local and the central dynamics, however, as it contains abundant data contents and updates fast, how to efficiently classify, organize and searches conference videos effectively has become an important research field. Due to the rapid development of multimedia data searching technologies, traditional retrieval technology cannot meet the retrieval user's demand of the variety and complexity of multimedia data. In light of this situation, more and more scholars turn to research the retrieval technology by automatic feature extraction to retrieval video data. Content-based video retrieval (CBVR) system is a system which structures the video data and uses the still features and motion features to retrieve the wanted videos by similarity degree. CBVR means video retrieval from enormous video database through matching of patterns derived from video analysis and feature extraction to represent its content. And it combines the theories and technologies of database, artificial intelligence, image recognition, and DISA, whereas, it has become a focus on retrieval technology.Firstly, this paper studies domestic and overseas present status of existed video retrieval systems. A detailed investigation and analysis about characteristic of video data is provided, and a model of video hierarchical structure is established. Based on referring and reading a large amount of documents for reference, this paper discusses arithmetic of key technologies such as shot detection, key-frame extraction and scene clustering in detail, and arithmetic are compared in order to discern their similarities and differences. Secondly, according to characteristics of party-government conference videos, a massive database is implemented for system functions, arithmetic of shot detection and key-frame extraction are discussed, principles of spokesman detection and speech recognition used in system are analyzed. Finally, this paper designs a simple system which consists of indexing and retrieval modules. Video indexing model includes functions of video structure analysis and feature exaction. Retrieval module is composed of user query interface, matching filter, candidates output, the adjusting feature for feedback and retrieval results. Through the analysis of experiments results, it is proved that this system finally obtained satisfactory retrieval effect.